Ways to solve complaining about ATI's filtering?

[Pete]

I think Damage's conclusion was warranted: whereas he previously gave ATi a "pass" when it came to IQ, this bit of subterfuge levels the playing field of suspicion. Now, as to whether trylinear is a bad thing, we've yet to find out, though I'm leaning toward looking at it as a good thing. But the very fact that ATi didn't tout is seems less than encouraging for the average frazzled web reviewer, no? Part of the reason everyone freaked over nV's 3DM optimizations is b/c the very notion of such a under-the-radar optimization requires far greater vigilance (and more understanding when it comes to interpreting benchmark numbers).

I'd love for someone with Mafia and a 9600 to test that road image with bi and try, and tri and bri. Ante P must have the two around. Let's see if he reads this.

 

[Quitch]

The fact that only 50% (less now) can't find a difference should give you the answer. If the quality was the same, i would say that at least 80% of the voters wouldn't see it.
Note that ~3/4 of the other voters did find the "good" answer.

But DaveB has gotten a copy of a program which shows legacy-trilinear and trylinear filtering, and the blind test said that most of us couldn't see any diff...

I disagree with the bold part.

Ok, that is your choice, but why?
Currently the poll is at 46% "Can't Tell"/"see diff, but can't decide which is correct", with 39% saying"Right" and 8% saying "Left"
And when I posted the post you quoted, the % for "Can't Tell" was even higher (51% I believe).

Thus I think that the majority still has trouble determining which is correct.
Alas the answer has been posted in the thread, so the results are getting "tainted" since we can not determine if the vote was based on what one saw or what one read in the thread.

So if you'd care to elaborate why you disagree I'd appreciate it

Woah there, Evildeus. The poll was "Which side is trilinear" not "Which side is better". Take thy abuse of the word quality else where

 

[Evildeus]

Of course, but isn't trilinear supposed to give better quality between 2 different mipmaps? Otherwise why do it? If they are able to spot the full trilinear, it's because the other one is not doing something? What could it be? Oh yeah a visual of the step between each mipmap? Meaning a diminussion of quality don't you think?

I for myself didn't search for global better quality but for visual mipmaps discrepancies. Unfortunately, If you look at the different post of the actual poll thread, you will see that most people looked @ the quality of the pic not for mipmaps discrepancies.

I do like the post made by ingenu in the poll thread.

I can see the difference and find the "correct" trilinear on a LCD screen.

The only problem I have is that I find the left part better than the right...

 

[Quitch]

Because between bilinear and trilinear, trilinear works in every situation and makes mip map level divides practically invisible. It does not lead to the best texture quality as the divides blur the textures, and thus when you can use less filtering, you should do so. The problem was that until now you used a global setting (unless the developer specified otherwise) and thus trilinear was the best bet since, while you lost a little sharpness, you removed the far more annoying mip map divides.

Trilinear is not, however, God's gift to IQ. It is filtering designed to remove an effect introduced by mip maps, while itself having its own side effects. ATI's method, if done well enough, would establish a far better method of determining the level of filtering where the developer has not specified otherwise. Just like the control panel, global settings are never the best option.

 

[Evildeus]

Did i say otherwise?

 

[digitalwanderer]

love for someone with Mafia and a 9600 to test that road image with bi and try, and tri and bri. Ante P must have the two around. Let's see if he reads this.

Ooooh! That'd be a great game to compare it on, the road really would give it away.

I'll hunt around for me Mafia disks, it's been ages since I played it anyways and it's a great game!

 

[Quitch]

Did i say otherwise?

You said "Isn't trilinear supposed to give better quality between 2 different mip maps". No. Trilinear is supposed to hide the divide between two mip maps. The less filtering used to hide that divide the better.

 

[Evildeus]

Did i say otherwise?

You said "Isn't trilinear supposed to give better quality between 2 different mip maps". No. Trilinear is supposed to hide the divide between two mip maps.

Yeah that's what i meant, sorry if my english is not as good as yours. But i'm sure if you wanted to understand what i said you would have. For example, in the same paragraph, i said "visual of the step between each mipmap" which is what trilinear is supposed to hide.

The less filtering used to hide that divide the better.

Yeah, but i can see the steps (with 40% of people), so what? A quick view is enough to spot the differencies on DB's pic (middle left/right).

 

[Althornin]

The less filtering used to hide that divide the better.

Yeah, but i can see the steps (with 40% of people), so what? A quick view is enough to spot the differencies on DB's pic (middle left/right).

no, 40% of people can see a DIFFERENCE...
not see the steps.

 

[nelg]

I am beginning to think that ATI should just re-badge their method of trilinear to something like “optilinearâ€￾ and drop any reference to trilinear in the drivers.

 

[Evildeus]

The less filtering used to hide that divide the better.

Yeah, but i can see the steps (with 40% of people), so what? A quick view is enough to spot the differencies on DB's pic (middle left/right).

no, 40% of people can see a DIFFERENCE...
not see the steps.

And what is the difference when we are talking of trilinear filtering? If it's not that, then it's not worth a dam

 

[MrGaribaldi]

The less filtering used to hide that divide the better.

Yeah, but i can see the steps (with 40% of people), so what? A quick view is enough to spot the differencies on DB's pic (middle left/right).

no, 40% of people can see a DIFFERENCE...
not see the steps.

And what is the difference when we are talking of trilinear filtering? If it's not that, then it's not worth a dam

I for myself didn't search for global better quality but for visual mipmaps discrepancies. Unfortunately, If you look at the different post of the actual poll thread, you will see that most people looked @ the quality of the pic not for mipmaps discrepancies.

(My emphasis)

Thus couldn't it be that people were looking at which side had the most blur?

Granted, that's not what trilinear, usually is, used for (rather it's a side effect), but it's still a difference from a texture not using a trilinear filter...

 

[Evildeus]

That's why i asked for a precision of what trilinear does. If you don't know, how can you compare? But well, doesn't matter.

 

[Althornin]

no, 40% of people can see a DIFFERENCE...
not see the steps.

And what is the difference when we are talking of trilinear filtering? If it's not that, then it's not worth a dam

the difference is, seeing a noticeable (barely) difference DOES NOT MEAN the steps are more or less visible, just that there is a difference!

This was a test to see if you could see a difference. You are interpreting that to mean: 50% of the people can tell (and thought) that "regular" trilinear looks better at mipmap translations.

That is not what the poll measured - you are incorrect.

People know what trilinear looks like. People know that pushing the mip map translations further back means that ATI's "trylinear" is in use, because the transition to a different mip level will start later.
Note that this means they could see the transition in BOTH HALVES of the image. this poll is not a quality judgement, which is what you are trying to amke it out to be.
Sorry, you are WRONG.

 

[Evildeus]

Which side is "Traditional Trilinear"?

If people know what to look for and can spot the true trilinear, it's that Ati's optimisation is doing a poorer work in mipmaps translation, hence less quality in mipmaps translation. Otherwise you can't spot it.

It's a quality judment on a particular effect, weither you like it or not.

That is not what the poll measured - you are incorrect.

In your immense knowledge, could you tell my humble person, what it does mesure?

 

[Socos]

You guys dont get it. ATI will do everything possible not to allow for a full trilinear option because of the not so wonderful benchmark results it will produce. This is because most sites will start benchmarking with trilinear set to ON and so the purpose of the new optimizations will be defeated. Benchmarks sell a product - like it or not. Thats why it is called benchmarketing. ATI and nVIDIA are out on the market to make money and sell products. Benchmarks sale products. End of story.

EHHHHhhh? Why would the benchmark results be so poor? The 9800XT does full TRI and it was no slouch just a month ago, so are you saying with full TRI that the X800 would perform less than a 9800XT? Or are you saying that it would just perform slower than a 6800?

Either way it doesn't matter too me if they "Optimize" I just want the option to be able to turn it on and off. Personally with everything I've seen I probably would always choose to have it on, but at least I could choose.

 

[Althornin]

Which side is "Traditional Trilinear"?

If people know what to look for and can spot the true trilinear, it's that Ati's optimisation is doing a poorer work in mipmaps translation, hence less quality in mipmaps translation. Otherwise you can't spot it.

WRONG.
Damn it, READ. READ. READ.
I'm sick of telling you over and over again.
But, one more time: The poll DOES NOT SAY "Which looks better". You equate being able to differentiate between two filters with a quality judgement between them, and that is simply NOT SO.

It's a quality judment on a particular effect, weither you like it or not.

That is not what the poll measured - you are incorrect.

In your immense knowledge, could you tell my humble person, what it does mesure?

I already did, multiple times.
i suggest you READ.
IT IS NOT A QUALITY JUDGEMENT.
If it were, it would have been titled:

Which side looks better?

Now, please read me previous post, and respond to the points theirin - do not simply ignore them, thanks.

 

[WaltC]

I think Damage's conclusion was warranted: whereas he previously gave ATi a "pass" when it came to IQ, this bit of subterfuge levels the playing field of suspicion. Now, as to whether trylinear is a bad thing, we've yet to find out, though I'm leaning toward looking at it as a good thing. But the very fact that ATi didn't tout is seems less than encouraging for the average frazzled web reviewer, no? Part of the reason everyone freaked over nV's 3DM optimizations is b/c the very notion of such a under-the-radar optimization requires far greater vigilance (and more understanding when it comes to interpreting benchmark numbers).

I'd love for someone with Mafia and a 9600 to test that road image with bi and try, and tri and bri. Ante P must have the two around. Let's see if he reads this.

Where TR really missed the boat, though, was here:

...Encouraging reviewers to make comparisons to NVIDIA products with NVIDIA's similar trilinear optimizations turned off compounded the offense. Any points ATI has scored on NVIDIA over the past couple of years as NVIDIA has been caught in driver "optimizations" and the like are, in my book, wiped out.

This simply makes no sense, considering that it's been plain for sometime that ATi's optimizations automatically shut off when running colored mipmap testing software, such as UT2K3/4, or 3D Analyze, such as ATi recommends to reviewers to use when comparing nV40 and R4x0 on the basis of trilinear IQ.

TR apparently is incensed because it thinks ATi was running colored mipmap software with its R4x0 Catalyst optimizations on while advising reviewers to turn off the corresponding nVidia optimizations--which indicates to me that TR is confused about how the optimizations function in the Catalysts. When running colored mipmap software the Catalysts automatically switch off the optimization and revert to "legacy" trilinear, and so when comparing nV40 the only way to get apples-to-apples inside of colored mipmap software is to turn off the nVidia optimizations manually, just as ATi advises in the documents the TR article highlights. That way both R4x0 and nV40 are running the colored mipmap software with "legacy" trilinear filtering, and neither is employing any trilinear optimizations (ostensibly.)

This is the most egregious error I've seen TR make in awhile, and I think it may be the automatic on-off nature of the Catalyst optimizations, versus the manual on-off switch found in the Forcenators, that has confused TR into thinking that just because there's no manual on-off switch that the Catalyst optimization is always therefore "on"--when in fact it's a self-switching proposition in the Catalysts, and switches itself off without user intervention in the event it encounters conditions that the Catalyst programmers feel are better served with "legacy" trilinear, such as colored mipmap software, or otherwise. As ATi puts it:

...If it were the case that we were only performing full trilinear with coloured mipmaps then you might have a point, but this is emphatically not what we do. We take advantage of properties of texture maps for performance and IQ gains. In cases where we are not able to determine that the texture content is appropriate for these techniques we use legacy trilinear filtering. This includes cases such as dynamically uploaded texture maps where we avoid performing analysis so as not to cause any possible gameplay hitches.

Eg, it's not apparent to me that TR realizes that the trilinear optimizing algorithm invokes both full trilinear and optimized trilinear support as per the software conditions programmed into the algorithm--so it may be either on or off, depending on the software, without any user intervention required.

 

[Rolf N]

Because between bilinear and trilinear, trilinear works in every situation and makes mip map level divides practically invisible. It does not lead to the best texture quality as the divides blur the textures, and thus when you can use less filtering, you should do so. The problem was that until now you used a global setting (unless the developer specified otherwise) and thus trilinear was the best bet since, while you lost a little sharpness, you removed the far more annoying mip map divides.

Trilinear is not, however, God's gift to IQ. It is filtering designed to remove an effect introduced by mip maps, while itself having its own side effects. ATI's method, if done well enough, would establish a far better method of determining the level of filtering where the developer has not specified otherwise. Just like the control panel, global settings are never the best option.

Trilinear does never produce blurrier textures than bilinear filtering. If you think it does, you need to do some more testing.

 

[Colourless]

Actually it does, because it will begin to use lower mip levels earlier.